Fleet Utilization Measurement: Why Reassignment Latency is the Gap Nobody Attributes

Reassignment latency is the time a vehicle spends available but unassigned, most often between completing one route and being given the next. It is a component of the fleet utilization gap that telematics cannot detect, because the engine is off and the vehicle is stationary at a depot, which reads as an idle asset rather than as a dispatch delay. Utilization reporting tells you the gap exists. Only dispatch data tells you how much of it is this.

Why the Utilization Number is Not the Problem

Fleet utilization measurement is well established and the standard definitions are sound. What they do not do is attribute the shortfall. A fleet reporting 78% has 22% of available vehicle hours going somewhere, and the question that decides whether anything improves is which component that 22% is made of, because each component has a different owner and a different fix.

Telematics attributes one part of it precisely. Engine-on idling, warm-up, queueing at a gate and stationary time with the engine running are all detected, quantified and reported, and they are legitimately worth managing. Fuel is the visible cost and the tooling for it is mature.

What telematics cannot attribute is the part where the engine is off. A vehicle back at the depot at 14:10 having finished its route, with its next assignment issued at 14:50, is not idling in any sense a telematics platform recognizes. It is parked. Nothing in the vehicle knows the difference between parked because there is no work and parked because the work exists but has not been assigned.

Cost pressure makes the distinction worth resolving. ATRI’s operational costs of trucking analysis put the industry average at $2.336 per mile in 2025, the highest in the report’s history, with costs up across every major line item. When the marginal cost of a mile is rising, the return on making an already-paid-for vehicle productive for another 40 minutes a shift rises with it.

Congestion consumes the same shift from the other end. The INRIX 2025 Global Traffic Scorecard recorded a US average of 49 hours lost per driver, rising to 102 in New York and 112 in Chicago. Time lost to traffic is largely outside the operator’s control. Time lost to reassignment is entirely inside it, which makes it the cheaper of the two to attack.

There is also a reason this particular component stays unattributed, and it is structural. Reassignment latency is produced by the interval between systems rather than inside any one of them, and Gartner’s survey conducted in October and November 2025 found that more than half of chief supply chain officers, 56% of those surveyed, cite integrating AI with legacy systems and processes as a major challenge. A gap that belongs to the handoff between systems has no natural owner in the reporting stack.

The general form of this problem is well documented outside logistics. MIT Sloan Management Review puts the cost of bad data at 15% to 25% of revenue for most companies, and notes that roughly two thirds of those costs can be identified and eliminated once they are located. An unattributed utilization gap is the same shape of problem: the loss is already being paid, and the work is finding out what it consists of rather than authorizing new spend.

Also Read: What Is Fleet Utilization? Key Metrics and Importance in 2026

How to Attribute the Utilization Gap

1. Fix the denominator first

Three denominators are in common use and they answer different questions. Total calendar hours measures capital productivity. Scheduled shift hours measures daily operational performance. Available vehicle hours, which excludes maintenance and compliance downtime, is the one that isolates dispatch performance, and it is the correct denominator for this analysis. Everything below uses an 8-hour shift of available vehicle hours, so 480 minutes with maintenance and compliance already removed.

2. Decompose the residual rather than reporting it

At 78% utilization on a 480-minute available shift, route time is roughly 374 minutes and the residual is 106 minutes. That residual is the whole subject. Split it into four components: engine-on idle at stops and in traffic, pre-trip and post-trip activity, empty repositioning, and reassignment latency. The first three are measurable with existing tooling. The fourth requires dispatch logs.

3. Calculate reassignment latency directly

Reassignment loss per shift is the number of routes a vehicle completes multiplied by the average latency between completion and next assignment. It is a two-variable calculation and both variables are in your dispatch data.

The sensitivity is worth internalizing. Each minute of average latency costs 0.62% of shift utilization at three routes a shift and 1.25% at six. Route count is set by network design and is hard to change. Latency is set by the dispatch layer and is not.

4. Compare it against the gap you are trying to explain

This is where the argument becomes concrete. Take the common case of four routes a shift with 20 minutes of average reassignment latency. That is 80 minutes, or 16.7% of the shift, against a total utilization gap of 106 minutes.

Reassignment latency is therefore about 75% of the entire gap, leaving 26 minutes to be shared between engine idle, pre-trip and post-trip activity and empty repositioning. An operation working hard on engine idle in that situation is optimizing the smaller quarter.

5. Model what closing it is worth

Hold the other 26 minutes fixed and vary only latency.

6. Convert the utilization gain into vehicles

Utilization points are abstract. Vehicles are not. Cutting average latency from 20 minutes to 10 raises effective capacity by a factor of 1.106, which means the same daily output from roughly 90 vehicles per 100 running today. At 5 minutes it is 86 per 100. That is the number to put in front of a fleet budget, because it is a capacity increase that requires no acquisition, no financing and no additional driver hiring.

Two cautions on reading it. This is shift-level time utilization for a deployed vehicle, not fleet-wide utilization across a full week, which is separately capped by the ratio of peak to average demand and by mandated non-driving time. And the modelled improvement assumes the work exists to be assigned. Where it does not, the correct conclusion is that the fleet is oversized for the demand rather than badly dispatched.

7. Segment before acting

Fleet averages hide the fixable cases. Segment latency by depot, by time of day and by wave. Latency concentrated in a single depot points to a local dispatch practice. Latency concentrated in the early afternoon points to a batching threshold waiting to accumulate stops. Latency spread evenly points to a system-level reassignment cycle, which is the most expensive to fix and the most valuable. Run the segmentation before proposing a solution, because the three findings lead to three completely different interventions and only one of them involves the platform.

Also Read: AI-Powered Fleet Utilization Analytics 2026

Engine Idle and Reassignment Latency Compared

The last row is the strategic point and the reason this metric is worth owning. Every major fleet telematics platform can detect and reduce engine idle, because it is a vehicle-state problem and vehicle state is what those platforms observe. None of them can reduce reassignment latency, because the cause is not in the vehicle. It is in the interval between a route completing and a decision being made, which lives in the dispatch layer.

Consider two fleets reporting the same engine-idle percentage. Fleet A runs three routes a shift with 8 minutes of average latency, losing 5% of the shift. Fleet B runs five routes with 32 minutes, losing 33%. Their telematics dashboards can look identical, because in both cases the engine is off while the vehicle waits. Any comparison of these two fleets on telematics data alone is measuring the wrong variable.

This also explains why fleet utilization benchmarks published in vendor roundups are close to unusable. Two operations quoting the same figure can have entirely different amounts of recoverable capacity inside it, and the number that would distinguish them is not one either operation currently reports.

Five Things to Instrument

1. Time from route completion to next assignment, p50 and p95. The p95 matters more than the mean, because a small number of long waits does most of the damage and averages conceal them.

2. Assignment cadence. How often the system evaluates unassigned work against available vehicles. A 30-minute planning cycle guarantees an average of 15 minutes of latency before any other factor is considered.

3. Batching threshold and the wait it creates. If dispatch holds orders to accumulate stop density, measure the wait separately from the density gain. Both are real and the trade is worth pricing rather than assuming.

4. Reassignment latency by depot and time band. This is the segmentation that turns a fleet number into a work list, and the pair of dimensions is what identifies where to intervene.

5. Share of the utilization gap attributed. Track what percentage of your 22% has a named cause. Most operations can attribute the engine-on portion and stop there, which means the majority of the gap is uncategorized and therefore unmanaged. Set a target for attribution coverage rather than for utilization itself in the first quarter, since you cannot move a number whose composition you do not know.

Also Read: Fleet Utilization for 3PLs: Multi-Client, Multi-Fleet Operations

What This Looks Like in Practice

Planning cycle time as the direct lever. Locus customers connecting warehouse readiness signals to automated dispatch have reduced planning cycle time by 66%. Planning cycle time and reassignment latency are closely coupled, because a vehicle waiting for the next planning run inherits the length of that run. Compressing the cycle compresses the wait, which is why this is the single most transferable number in the analysis. It is also the number worth asking any vendor for, since a platform that cannot state its assignment cadence cannot claim an effect on utilization.

Execution rate at fleet scale. A Fortune 50 operation running more than 4,500 drivers moved execution rate from 75% to 92% and surfaced more than $14M in annualized operational opportunity. A 17-point improvement of that kind is not produced by observing vehicles more closely. It comes from reducing the interval between a change in state and a decision about it.

Consolidation as the enabling step. A retail enterprise consolidated six legacy systems into a single execution layer, cut manual dispatch effort by more than 80%, held 99%+ on-time delivery and broke even inside year one on $1M+ in savings. Reassignment latency is usually a handoff artifact, so removing handoffs removes latency without anyone tuning a threshold.

Four Measurement Mistakes to Avoid

Reporting the gap without decomposing it. A utilization percentage is a symptom. Without attribution it produces a target rather than a work list, and targets without causes get met by changing the denominator.

Treating all stationary time as one category. Engine-off waiting and engine-on idling have different causes, different owners and different fixes. Collapsing them means the larger one is managed with tools built for the smaller one.

Benchmarking utilization against other fleets. Route count per shift varies enormously by operation, and the same latency produces very different utilization loss at two routes than at five. Compare against your own segmented baseline instead.

Assuming latency is a dispatcher discipline problem. Most of it is set by assignment cadence and batching thresholds, which are configuration rather than behavior. Coaching a dispatcher will not shorten a 30-minute planning cycle. The test is whether latency falls at night or at weekends when the same people are not on shift. If it does not move, the cause is configuration rather than practice.

Also Read: Fleet Management and Utilization: AI Architecture Framework 2026

How Locus Reduces Reassignment Latency

Locus, the world’s first Decision-Intelligent, Agentic TMS, is built around continuous reassignment rather than scheduled planning runs, which is the property that determines reassignment latency. A vehicle completing a route is a change in state, and a system that re-solves on state change assigns the next piece of work as an event rather than at the next cycle. That is the mechanism behind the 66% reduction in planning cycle time, and it is the same mechanism that closes the gap this analysis quantifies.

Within the DiSCO framework the Capacity and Dispatch Agents run a continuous Sense-Decide-Execute-Learn loop, with partial replan so recoverable work is reassigned without disturbing routes that are executing correctly. Dynamic batching evaluates the trade between stop density and waiting rather than holding to a fixed threshold, and capacity blending across owned, third-party and gig fleets means the next assignment can come from a different supply pool when the nearest available vehicle is not the right one.

The platform models 250+ real-world constraints simultaneously, which matters here because a fast reassignment is only useful if it is also a legal and feasible one. Assigning work in seconds to a vehicle that cannot carry it, or to a driver at their hours limit, converts a latency problem into a rework problem.

Locus runs at 1.5B+ deliveries across 360+ enterprise customers in 30+ countries at 99.99% uptime, with DispatchIQ sustaining 99.5% on-time delivery against the 80% to 90% typical of manual dispatch. Locus has been recognized by Gartner for seven consecutive years across multiple research categories, appears in the 2026 Gartner Hype Cycle for AI-powered logistics, features ShipFlex as a Representative Vendor in the 2026 Gartner MCPMS Market Guide, holds Leader designation in the QKS SPARK Matrix for Transportation Management Systems, and ranks #1 on G2 for Route Planning software.

Gartner predicts that by 2031, 60% of supply chain disruptions will be resolved without human intervention. Reassignment is the least contentious category to move in that direction, because the decision is bounded, reversible and high frequency, which makes it the natural first candidate for autonomy rather than the last.

In October 2025, Ingka Investments, the investment arm of Ingka Group, the world’s largest IKEA retailer, acquired Locus. Locus continues to operate independently.

To measure reassignment latency across your own depots and time bands, schedule a demo.

Also Read: How to Reduce Fleet Idling and Save Fuel Costs in 2026